
Internet Protocol Television (IPTV) provides live TV channels to users over IP in the form of streaming. Since each channel is transmitted on request, there is always the problem of delay between channel changes. The channel switching delay is familiar problem in digital television and widely researched in the context of IPTV systems. Several solutions have been proposed to reduce this delay. A few popular solutions suggest transmitting more than one channel to users at the same time. However, choosing these channels is not trivial as the next channel a user will switch to is not known. In this paper, we analyze machine learning approaches to predict the next channel of the user. Such a prediction can be useful in design of frameworks for reducing the channel switching delay. We have performed our experiments on dataset collected from an operational system on which the Multilayer Perceptron(MLP) model produces better accuracy.